与目前最先进的语言模型不同,幼儿通过与周围环境和照顾者的互动来主动习得语言。一种被认为对语言学习至关重要的机制是推断社会环境中其他主体的心理状态的能力,由Premack和Woodruff(1978)创造的心理理论(ToM)。从Rabinowitz et al.(2018)和Zhu et al.(2021)实现的ToM的现代操作化版本中汲取灵感,我们构建了配备ToM的语言学习代理,并测量其对学习过程的影响。我们给说话者代理一个内部的听者模型,该模型与说话者一起训练,并用于对潜在的话语进行重新排序。我们用不同的任务难度进行实验,假设模型将获得更复杂的语言来适应更强的环境压力。我们发现,训练扬声器与一个高度加权的ToM听众组件导致性能增益在我们的图像参考游戏设置。我们还发现一些证据表明,在训练过程中增加任务难度会导致更流畅和准确的话语评价。这表明进一步将心理理论以及儿童语言习得的其他见解纳入语言习得的计算模型的潜在效用。
英文摘要:
Unlike current state-of-the-art language models, young children actively acquire language through interactions with their surrounding environment and caretakers. One mechanism that has been argued to be critical to language learning is the ability to infer the mental states of other agents in social environments, coined Theory of Mind (ToM) by Premack & Woodruff (1978). Drawing inspiration from the modern operationalized versions of ToM implemented in Rabinowitz et al. (2018) and Zhu et al. (2021), we build language-learning agents equipped with ToM, and measure its effects on the learning process. We model ToM by giving the speaker agent an internal listener model that is trained alongside the speaker and used to rerank potential utterances. We experiment with varying task difficulty, hypothesizing that models will acquire more complex language to adapt to stronger environmental pressures. We find that training speakers with a highly weighted ToM listener component leads to performance gains in our image referential game setting. We also find some evidence that increasing task difficulty in the training process results in more fluent and precise utterances in evaluation. This suggests the potential utility of further incorporating ToM, as well as other insights from child language acquisition, into computational models of language acquisition.
Psychomatics -- A Multidisciplinary Framework for Understanding Artificial Minds
Catholic University of Sacred Heart(圣心天主教大学);Istituto Auxologico Italiano IRCCS(意大利奥索洛吉科研究所);University of Milano - Bicocca(米兰比可卡大学);Virtual Reality Medical Center(虚拟现实医疗中心);Interactive Media Institute(交互媒体研究所)
A Knowledge-Based Language Model: Deducing Grammatical Knowledge in a Multi-Agent Language Acquisition Simulation
Leiden University Centre for Linguistics (LUCL), Leiden University, the Netherlands(莱顿大学语言学中心(LUCL)、莱顿大学、荷兰);Leiden Institute for Brain and Cognition (LIBC), Leiden University, the Netherlands(莱顿脑与认知研究所(LIBC)、莱顿大学、荷兰);City University of Hong Kong (CityU), Hong Kong(香港城市大学(CityU)、香港)
Unsupervised Acquisition of Discrete Grammatical Categories
Leiden University Centre for Linguistics (LUCL)(莱顿大学语言研究中心);Leiden University(莱顿大学);Leiden Institute for Brain and Cognition (LIBC)(莱顿脑与认知研究所);City University of Hong Kong (CityU)(香港城市大学)